Intuitions

This is the first time (I think) that I've ever used my posting privileges here, and since I don't know whether my preferences and name from my other blog will carry over here, I should say at the outset: this is Hilary Bok. I'm writing in response to a comment made by Kip Werking in an earlier thread; I'm writing a post about it since it seemed like the sort of topic that might justify it.

In the earlier thread, I wrote:

"consider an analogy. People have all sorts of conflicting intuitions about justice, and I assume this was as true in the late 60s as it is today, if not more so. What should John Rawls have taken this fact to entail? Probably: that just appealing to people's intuitions wasn't likely to get him anywhere. But it would not entail that there were no interesting, constructive arguments out there that might lead people to change their minds.

What bugs me about the current emphasis on intuitions etc. is that it seems to make the very possibility of these arguments vanish. If people's intuitions are all in accord, there's no problem to solve; if they differ, then it's unsolvable. In this way the entire constructive task of philosophy is made to disappear."

Kip asked whether I was talking about concepts or inferences, and said:

"If Hilary is saying that inferences are not a popularity content, then I completely agree with her. After all, the folk believe all sorts of ridiculous things. But it seems to me that they come to believe each of these ridiculous things through some fallacy: they have the right premises but reach the wrong conclusion. In that case, it is not a popularity contest. The folk are just wrong.

But if Hilary (and others) are saying that concepts are not a popularity content, then I beg to differ. I'm no philosopher of language (yet?), but it seems to be a bedrock principle of that subject, that terms means whatever their common usage is. And what is common usage but a popularity contest? You ask "which of these things is red" and point to some blood and some sky, if most of the folk say the blood, then that is the color "red", if most of them point to the sky, then "red" here actually means what we call "blue". This seems to be a fundamental truth about how language works. And it would be a most dangerous precedent to do violence to this principle just to (as it appears) inoculate some view from empirical investigation.

Given this distinction, the question is: is the compatibility question one of concepts or one of inferences? And it seems to me that it is much more one of concepts than of inferences. We are all bright, college educated people who know the simple rules of logic. And we know that, if "free will" means what G. Strawson says it means, then it doesn't exist, and if "free will" means what Daniel Dennett says it means, then it does exist (and is compatible with determinism). I don't know of a single fallacy that either side can point to the other, and accuse them of making. On the contrary, the only mistake either side could say is: "Sure, free will does/does-not exist if you use that concept of free will! but of course that is the wrong concept! You haven't committed any glaring fallacy, rather your argument never even got off the ground because you started with the wrong premise, the wrong concept of free will." And if this is true, and I believe it is, how can we settle this dispute, if not through empirical investigation, and determining what the common usage of "free will" is, and even confronting the possibility that it may have no common usage (as Richard Double seems to have concluded)! And if you appreciate how free will dispute is about concepts, and not inferenes, then you can understand why I tend to think that decisive answers about folk intuitions can be the "nails in the coffin" to compatibilism/incompatibilism."

Personally, I tend to the following views:

(a) I can, if I want, construct any concept I want. If I want to pick out that class of objects that consists solely of my house, Tom DeLay, and the number two, that's my business. It is of course true that I do not get to determine: whether my new concept actually applies to anything, and if so, what; what follows from the fact that it applies to something; etc. I also do not get to decide whether my concept is what most people mean by freedom (or by any other term.)

(b) What 'being wrong about a concept' means, I assume, is 'being wrong to think that I use the term 'X' to pick out the same concept as most other people', or something like that. To detect such an error, one would of course have to check and see what most other people do in fact mean.

(c) Personally, I am less interested in the question whether freedom, as most people use that term, is compatible with determinism, than in the questions: what is it that I want an account of freedom of the will to (allow me to) do? ('Do', here, isn't limited to actions; it includes things like: hold people morally responsible for their conduct, in some satisfactory sense. The parentheses are meant to indicate that you can construe this as a question about what the account can do -- e.g., can it underwrite an adequate account of responsibility? And yes, this means that I hold the same basic view on the problem of responsibility as on freedom: I need to ask what I want an account of moral responsibility to allow me to do, as well.) Why does freedom of the will seem to me, and to others, to matter? What turns on it? And is there some account of freedom that will allow me to get what I think turns on getting an adequate account of freedom?

(d) This means that I think of the problem of freedom of the will as the question whether it is possible to construct a concept that is both close enough to what we normally mean to count as a concept of freedom, and also able to do the various things we want an account of freedom to do. I take interesting versions of incompatibilism to assert that it is impossible to construct an account of freedom that is compatible with determinism/mechanism/naturalism/whatever, and that does what we want an account of freedom to do. (Since if they were only asserting that a given account of freedom is inconsistent with d/m/n/whatever, but allowing that there was some other account that gave us everything we wanted in an account of freedom, and which was compatible with d/m/n/w, their claim would not interest me.)

(e) It's hard to demonstrate this sort of impossibility -- to rule out the possibility that some new and unforeseen approach might actually succeed in letting us construct such an account. It's easy to gesture at the claim that such an account is possible, but it's pretty boring. What would be interesting would be to actually do the hard constructive work involved in coming up with one, which would of course demonstrate the possibility of doing so by, well, doing it.

***

Now: Kip asks whether I take my opponents to be wrong about concepts or inferences. There are, of course, people on both sides who I think have made mistakes of both kinds (I name no names ;) ) But the interesting mistake, I think, is the one I just discussed: thinking that it is impossible that any future argument will succeed where (assuming this for the sake of argument) all the existing ones have failed. This is not a mistake about concepts, but it's not obviously a mistake about inferences either -- at any rate, the 'mistake' would involve a mistake using a form of inference (induction) that no one takes to remove the possibility of error in any case.

Omitting the possibility of this sort of mistake is, I think, where Kip and I differ. And it's why I brought in Rawls to begin with. The topic of justice had been gone over pretty thoroughly when Rawls came along. Intuitions seemed to differ, and those differences seemed pretty intractable. I honestly think that had Rawls accepted, say, Richard Double's philosophical views, he would have concluded: darn, I guess people's views about justice are internally inconsistent, so there's no such thing as justice anyways. Time to move on to the next problem.

Luckily for all of us, he thought instead that there was room for constructive work: coming up with a different approach, working it out in detail, and arguing that it allowed us to reconcile things that had previously seemed irreconcilable, and to make tractable problems that had previously resisted any sort of solution. (You can play this game with lots of philosophers. Kant: "There are antinomies. So much for reason." Aristotle: "Let us consider the opinions of the many and the wise. They disagree. Darn.") And, as I said, what bothers me about the idea (Double's; I have no idea if it's Kip's) that when intuitions about some concept differ or are inconsistent with one another, we should just conclude that that concept fails to apply, is that it makes the entire possibility of doing interesting constructive work vanish. Because interesting, constructive work is what you try to do when you think that all the obvious paths have been tried and some entirely new approach is called for. This doesn't require thinking that your opponents have made mistakes either about concepts or about inferences. It only requires thinking that not all the interesting approaches have yet been tried.

Comments

[Zeal]

You might be interested in this.

www.freewire.freespaces.com

Zeal

[Patricia Greenspan]

Hilary -- I quite agree with what I take to be your main point here, that a split in intuitions, even about concepts, doesn't mean that there's no truth of the matter. However, it's not obvious that the truth has to involve a new, "constructive" approach. For instance, intuitions about concepts could be inconsistent, with people saying one thing when asked about freedom in general terms and something else when asked about cases -- the sort of thing Rawls meant for us to minimize of by working toward "reflective equilibrium." Note that Rawls's constructive theory of justice by no means got rid of the division in popular intuitions on the matter; I guess his claim would be that reflective equilibrium should accomplish this, with his theory as a further option on the general level. Somehow, though, I think disagreement would persist -- along with meta-disagreement about whether people on the other side were properly engaging in reflective equilibrium.

In any case, when you describe your own views, I have a problem with reinterpreting a concept like freedom of the will just in terms of something one wants to do with it, i.e. hold people morally responsible, taking anything else that might underlie many people's intuitions as irrelevant. This is a common move these days, not just yours. I note that you shift quite readily between talk in (c) of what "I" want to talk in (d) of what "we" want an account of freedom to do. But it would be unfortunate to interpret that "we" as referring to anyone whose intuitions should be taken seriously, dismissing those who have no particular (or equally worthy) *job* in mind for the concept they mean to be employing. Come to think of it, perhaps another "constructive" role for philosophy might be to expand our notion of the roles such a concept might play -- e.g. in an individual's self-conception, apart from the kinds of moral applications that philosophers need a toolkit for.

Incidentally, it occurs to me to ask whether you think that Rawls, in coming up with a new account of justice, does something comparable to what I'm questioning. Would he say, of accounts of justice as something other than fairness (Aristotle on proportionality to merit comes to mind), "they don't interest me" -- not just in the sense that they're not what he's trying to give an account of, but also that nothing turns on them (as you seem to be saying for freedom-apart-from-responsibility)? I don't mean this question as rhetorical; maybe he *would* say something of the sort.

[RICHARD DOUBLE]

[1] Like Hilary Bok, I see no philosophical [rather than psychological] value in consulting intuitions based on popular usage for deciding, e.g., compatibility questions. I emphasize that the conflicting intuitions I think important are the best ones experts reach upon serious reflection. [2] I do not deny that philosophers profitably can construct concepts they find useful for specific purposes. I deny that such constructions tell us the truth about whether, e.g., retribution is really justified or not. [3] A more apt comparison to my view is to conclude from Kant’s antinomies that space and time are subjective, not to reject reason.

[hilzoy]

Patricia: I was probably unclear. First, about I and we: mostly, I was just using 'I' as an instance of 'we' in the post. (Or rather: this is true when I'm talking about what we/I think it important that a concept of freedom allow us to do. When I say something like: I wouldn't bother working on freedom of the will if it didn't seem to me to matter to anyone, then 'I' is (specifically) me.) In any case, I think that if I want anything I write to be of any interest or use to anyone else, I should try to figure out what about it might fail to satisfy them.

Second: I meant underwriting an adequate conception of responsibility to be an example of one of the things we want an account of freedom to do, not the only thing we want it to do. So my point wasn't meant to be 'anything that lets me hold people responsible has met all my criteria', so much as: let's try to come up with all the reasons why we think it matters that we be free. (I completely agree that things like 'being important to one's self-conception' should count here, though I also think it's worthwhile to figure out why something matters in that way. -- One of the things that always bothered me about early, Ayer-like versions of compatibilism was that they seemed to omit this entirely.) Suppose some concept X does this: i.e., if we.are X, then we get all we think matters in the free will debate, and (moreover) we are X. What then?

Well: X might be quite unrelated to anything we'd normally call 'freedom'. Suppose all that good stuff turned out to follow from our being bipeds: I wouldn't say that freedom meant being a biped. That's why I said (above) that I was looking for a concept that's "close enough to what we normally mean to count as a concept of freedom". If it is such a concept, and if (if we.are X, then we get all we think matters in the free will debate, and (moreover) we are X), then I'm satisfied.

About constructive approaches: here I meant something pretty basic (not, for instance, having anything to do with constructivism.) What bothered me about the question Kip posed was that it seemed to assume that all the arguments were already there, laid out, so that we just had to decide whether our opponents were wrong on (a) inferences or (b) intuitions. What bothers me further about Richard Double's approach is that it seems to reduce philosophy to noting agreement on intuitions and their application or concluding that the concepts involved have no application, thereby eliminating all the interesting work of trying to put one's own view together (which is basically what I meant by 'constructive work'.)

[Patricia Greenspan]

Hilary -- Thanks for the clarification. I do see what you mean on these points. But am I right in thinking that a "constructive" approach, as you're taking it, would mean coming up with a new view? Though it's one good response, I'm not sure why it would be required by a persistent clash in intuitions, particularly since it might just introduce another clash. Explaining what different conceptions underlie the clash -- and in some way assessing them and arguing in favor of one over the other -- would seem to be the way to defend a particular view on the issue, old or new. It's the thought that intuitions on the unanalyzed concept in question are "rock-bottom" that should be questioned. Perhaps this broader point is the one you're really making.

Incidentally, on the question of online id's: though I don't mind "Patricia," I'm usually called "Pat." Didn't think of that when signing up, and I'm inferring from your own situation that it's too late to change it.

[Tamler Sommers]

Hilary,

You write "This means that I think of the problem of freedom of the will as the question whether it is possible to construct a concept that is both close enough to what we normally mean to count as a concept of freedom, and also able to do the various things we want an account of freedom to do."

How are we to find out (a) what we normally mean by the concept of freedom, and (b) what we want an account of freedom to do, if we don't engage in a (sophisticated) investigation of intuitions regarding (a) and (b)? It seems to me that placing some value in this type of investigation is fully compatible (and even complementary) to the Rawlsian constructive approach you want to promote.

I don't see how probing people's intuitions makes the possibility of argument or even of changing people's intuitions vanish in any way. On the contrary, I would think it could lead to new and more developed arguments and theories. Even in my post, where I assumed for the sake of argument that intuitions fundamentally clashed on the question of desert, I wasn't ruling out the possibility that further argument could be made. I was just asking what kind of argument it would be, since it could no longer appeal to our allegedly incompatibilist intuitions. But I was fully open to the possibility that the argument might a good or helpful one.

[Kip Werking]

Well, I was at once flattered and frightened (a little of both) at seeing an entire post, and not just a comment, in response to my last comment about intuitions.

As one of the less succinct Gardeners, I have several points to make:

1. “To detect such an error, one would of course have to check and see what most other people do in fact mean.”

With this sentence, Hilary seems to agree with me. But it is not clear how this sentence fits with the other remarks she has made about investigating intuitions. Is checking for this error not interesting philosophical work? Should we not check for this error if it threatens to make “interesting constructive work” vanish?

Suppose, as Pereboom, G. Strawson, and I think, that this error explains a major view in the contemporary debate about free will: compatibilism. This would seem to make checking for this error interesting and important philosophical work. So it is not clear to me why Hilary seems so hostile to people actually doing this work (beside the obvious fact that she favors compatibilism!).

2. The conceptual danger I am talking about is nothing new to non-realists or eliminativists about free will. Derk Pereboom anticipates Dennett in LWFW:

“His attempt to recast the debate in terms of the question ‘What is free will such that we should want it?’ potentially confuses two issues: Does we have the sort of free will required for moral responsibility? And do we want the sort of free will required for moral responsibility? It could be, for instance, that we are free in the sense required for moral responsibility, but since being free in this sense is not especially valuable to us, we would not want it much. It is important to frame the issue so as to make conceptual room for views of this type.” Living Without Free Will (2001) Cambridge: Cambridge University Press, at p. xxiv.

Similarly, Galen Strawson writes in his review of Dennett’s Freedom Evolves:

“Well, he's a ''compatibilist'': he thinks that freedom is wholly compatible with determinism, although determinism is the view that everything that happens in the universe is necessitated by what has already happened, so that nothing can ever occur otherwise than it actually does. He thinks, in other words, that you can be wholly free and morally responsible for your choices and actions even if every single one of them was determined by events that happened long before your birth. You think this a strange notion of freedom? Me too.”

So, I don’t want my warnings about this conceptual danger to seem too idiosyncratic. The handful of philosophers who actually deny the existence of free will seem to have been pressing this point for years.

3. To make this point, I’ve personally suggested the following analogy:

“Unicorns don’t exist. But suppose that belief in unicorns was one of the most precious beliefs that human beings have about the world. Suppose that many or most people tend to believe in unicorns uncritically, without reflection, like many believe in witches or young-earth Creationism. Then more reflective philosophers, in a marketplace of ideas, came to confront their belief in unicorns. They inspected the world and discovered, to their astonishment, that unicorns do not exist. Then the most remarkable thing happened: instead of giving up their precious belief in unicorns, they searched for something similar. They saw that although unicorns do not exist, horses do. And so they called horses unicorns. And then when all of the naïve people asked them ‘do unicorns exist’, they could answer, like the triumphant defenders of all that is good and sacred: ‘yes, they do exist.’ But if the masses knew that the philosophers only meant ‘horses exist’ they would have felt disappointed and angry.”

3. And I would make another point: just as Kant said “existence is not a predicate”, I would say that desirability is not a predicate. This is implicit in the Pereboom excerpt. One can imagine the most horrible thing, e.g. the Holocaust, or the most wonderful thing, e.g. Heaven, but whether or not each of those things is desirable is a further question, beyond the question of what each thing is. Therefore all talk of desirability, or “the kinds of free will worth wanting” seems misplaced. It jumps the gun. As Pereboom says, let’s decide whether free will exists first, and then we can decide whether it is desirable.

4. Now, Hilary anticipates this argument:

“Personally, I am less interested in the question whether freedom, as most people use that term, is compatible with determinism, than in the questions: what is it that I want an account of freedom of the will to (allow me to) do?... Why does freedom of the will seem to me, and to others, to matter? What turns on it? And is there some account of freedom that will allow me to get what I think turns on getting an adequate account of freedom?”

The first sentence implies that, for all Hilary is interested in, “free will” according to the folk does not exist, but “free will” according to Hilary does exist and accomplishes her purposes. I think that is a fair reading of her comment. Consequentially, we don’t need to put much emphasis on what the folk think free will is.

But as long as we avoid exploring folk intuitions, then one should be careful to name whatever other concept one uses as idiosyncratic. It would not be fair, given the fundamental truth that common usage dictates the meaning of terms, to talk about “free will” if you really meant something else (even something more desirable). One would need to talk about “Hilary-freedom” or “c-freedom” or some other thing (as Galen Strawson suggests). I do not know whether Hilary does this. Daniel Dennett certainly doesn’t.

5. There is also the danger of irrelevance. Hilary notes that “I can, if I want, construct any concept I want.” And of course she can. But philosophers have traditionally involved themselves with questions that seem relevant. “My house, Tom DeLay, and the number two” would not seem very relevant. Similarly, if one abandons the folk concept of “free will” and starts exploring other concepts, one must worry about the danger that the triumphant philosophers will come back and say “I found out that X exists” and the folk respond “so what? I don’t care about X. I care about Y.”

6. Let me add one note about why people find compatibilist concepts of free will too weak. Even if the non-realist/eliminativist grants the importance of desirability to the compatibilist (and we shouldn’t do this), compatibilist accounts of free will still do not pass muster. They don’t pass muster because, for example, they cannot ensure that one’s entire life story was not set up by a Grand Puppet Master like so many dominoes. The compatibilist accounts of free will that are allegedly “worth wanting” are actually still not good enough. That is the whole point of Mele’s Zygote Argument (amongst others) and the reason, or at least one reason, why he remains agnostic about compatibilism.

7. About induction: I certainly grant that induction produces no guarantees. I didn’t mean to imply otherwise. But perhaps the only thing more mistaken than thinking that induction conveys guarantees is to think that it provides no information whatsoever. Three thousand years of the traditional philosophical method has not produced a consensus about the solution to this problem. Maybe it’s time to borrow some tools from psychology and the sciences.

8. It would not follow, if Rawls had found a Double-ish argument against the existence of justice cogent, that he would or could do no interesting philosophical work, even philosophical work about a similar, but *distinct*, concept. He would just have to call it justice* or Rawls-justice or some other thing. Would that be such a tragedy? This is all that Pereboom, G. Stawson, myself and others ask of compatibilists who proffer concepts of “free will” that seem drastically too weak, and then, furthermore, seem to want to inoculate these weak concepts from empirical investigation. I think it is, in this respect, really a modest request.

Again, there is the danger of irrelevance: that nobody will care about Rawls’ idiosyncratic notion of justice, or that nobody will care about compatibilist notions of free will. But this danger seems remote, because even I find compatibilist notions of free will interesting. And doing this will have the happy consequence of clearing the conceptual muddy waters (so to speak), rendering our philosophical work more precise and accurate. And even if the consequences were dire, that would have no bearing on whether they are actual.

One consequence is inescapable: “free will”, as distinct from c-freedom, that thing that philosophers had argued about for ages, that thing that so many non-specialists had placed their hopes and faith in, will, at the end of the day, not exist. And the compatibilist might try to comfort the non-specialists and say “well, free will doesn’t exist, but c-freedom exists, and it’s just as good.” But the folk might, for reasons good and/or bad, not find too much solace in the compatibilist’s words.

9. Hilary repeats a point from her original post:

“And, as I said, what bothers me about the idea (Double's; I have no idea if it's Kip's) that when intuitions about some concept differ or are inconsistent with one another, we should just conclude that that concept fails to apply, is that it makes the entire possibility of doing interesting constructive work vanish.”

In response I would repeat my original counter: that whether something is good or bad has no bearing on whether it is actual. To think otherwise would be to commit what Carl Sagan called the Argument from Adverse Consequences. It would seem to indulge in wishful thinking (and note that wishful thinking is a prominent cognitive bias).

What I want to add now is this: even putting aside the worry about the improper influence of wishes and emotion, is this even a good wish to have? Let’s be clear, we are faced here, by hypothesis, with two possibilities: A. a quick and easy solution to one of the most vexing and profound problems to face the human species since the Greeks and B. the continuation of philosophical work on the problem, perhaps for another three thousand years, and preferably, or so it seems, in the empirical dark. Don’t get me wrong, I understand how B might be preferable to persons whose careers are based upon working on philosophical problems (by analogy, doctors have perverse incentives to keep their patients sick), and I love philosophical work as much as the next guy/gal, but I don’t love it at the expense of actually solving problems.

[Eddy Nahmias]

Sometimes people (like Kip above) suggest on this blog that what little work has been conducted so far on folk intuitions about free will/moral responsibility and determinism has shown that the folk have predominantly incompatibilist intuitions (at least when they are not emotionally jacked up). I don't think this is right, and I hope to be able to back up this claim with the current studies I am doing with Kvaran and Coates and the studies I will do with Tamler this summer (though he is hoping we get different results--that's why we're working with each other).

But even based on the current work (my co-authored initial studies) and Nichols and Knobe's work, I don't think the proper conclusion to draw is that people have predominantly incompatibilist intuitions. We tested the affect influence in our studies and found none, though all of our cases included concrete actions (people gave majority compatibilist-friendly responses across three different presentations of determinism and across morally good, bad, and neutral actions). Nichols and Knobe's scenarios, in my view, beg some important questions about how to present determinism and its relation to the ability to do otherwise (e.g., their scenario basically says that in the deterministic universe you can't do otherwise), so I don't think their results can be taken as disconfirming our earlier work, though I agree with them that affect and/or concreteness influences people's judgments.

Anyway, I just don't want people to get the impression that there is anything like a concensus or settled answer to the question of whether ordinary people are "natural incompatibilists."

[Kip Werking]

Eddy, you're absolutely right that the evidence is mixed at best. I hadn't thought I had implied otherwise, but reading back over what I've written, I can understand how some would get that impression. I just want to reiterate that my more thoughtful, and modest, position here is just that we should explore these empirical questions, wherever the experiments may take us, and that doing so is philosophically constructive, important, and interesting.

[Paul Torek]

Hilary, thanks for this excellent thread. I heartily agree with your commendation of a constructive approach. The objective should be to capture the interesting features that we associate with free will - ideally, all of those features. But possibly, some adequately large and important subset of those features suffices, if it is a larger and more important subset than any competing account.

But I do have a difficulty. It comes out when you write:

"(c) Personally, I am less interested in the question whether freedom, as most people use that term, is compatible with determinism, than in the questions: what is it that I want an account of freedom of the will to (allow me to) do? ('Do', here, isn't limited to actions; it includes things like: hold people morally responsible for their conduct, in some satisfactory sense. ... ) Why does freedom of the will seem to me, and to others, to matter? What turns on it? And is there some account of freedom that will allow me to get what I think turns on getting an adequate account of freedom?"

I guess my question amounts to this: by 'Do', here, how do you limit the scope of what you want in an account of free will? For example, say that I praise Fischer's account for capturing the intuition that current-day robots lack free will; or say that I condemn it for giving the 'wrong' answer to the Zygote Argument scenario. Is this something that an account can (fail to) 'do'? In that case the word 'do' seems not to do (sorry, can't help myself) any work. I don't think this is a good reading of what you're saying, though, so let me try another.

Perhaps you are thinking more along the following lines. (Let me put some words in your mouth, then you can spit them out and provide better.) "It might turn out that most people's linguistic intuitions necessarily connect 'free will' to things that just don't matter morally or in any other practical way. But in that case I, as a moral philosopher, am not interested in 'free will' conventionally understood, but rather in the qualities that underly moral responsibility, personal autonomy, ability to deliberate, etc. etc." Now if that's what you're saying, or something very like it, I agree with the spirit. I just have this quibble about "it might turn out that free will doesn't matter": I don't think that 'possibility' should be taken very seriously.

In my view, there really isn't much room for contrast between Kip's claim that philosophers should be true to what people normally mean by the term "free will", and Hilary's project of constructing an account that captures what's important to us about free will. That's because - despite something else Kip claims - any completely successful capturing of what's important to us about free will would be ipso facto a very successful capture of the ordinary meaning of the term "free will". Sure, there may be other linguistic intuitions we have about free will that pull us toward features with no practical importance. But if it comes to a contest between the important and unimportant features, the important ones will win out.

A concept can survive the discovery that some of the supposed features turn out not to exist. Dennett, if I recall, discusses the survival of 'mass' through the Einsteinian revolution, in Freedom Evolves. The question whether we've discovered that mass is different from what we had thought, or whether we've discovered that mass does not exist, amounts to whether the best candidate for 'mass' in an Einsteinian universe has enough of what was important about 'mass' to qualify.

Now admittedly, 'importance' in this context need not amount to importance in a moral sense, as the 'mass' example clearly demonstrates. But then, that's the difference between 'mass' and 'free will'. The latter is deeply entangled in our moral practices and perspectives. And, human nature being as social as it is, once we attach a concept to practices like deliberation, respect, autonomy, moral responsibility, etc., those features will dominate our thinking using the concept.

‘What is free will such that we should want it?’ is an unfair twist on Dennett. Rather, Dennett asks, "given that we know free will is worth wanting, what could it be?" And how do we know that free will is worth wanting? Here I leave Dennett behind, and answer for myself (although I have a slight suspicion he might agree). It's just another intuition. But not just another intution, it's a really big one. The payoff for abandoning it would have to be enormous. I should note that the same cannot be said for moral responsibility, which is part of the reason why it's not uncommon to drive a wedge between them, and go for some kind of semi-compatibilism (for example). It does help, though, if the wedge is a smallish one.

[Patricia Greenspan]

Paul -- May I ask for a clarification on your last suggestion, about responsibility vs. free will. Did you really mean to be denying that "the payoff for abandoning [moral responsibility] would have to be enormous"? Or is the point just that we could more readily modify our intuitions about it, allocating it on some other (e.g. utilitarian) basis?

Something else (though related to the question of driving a wedge between free will and moral responsibility): when we talk about what's more or less important, I think we really need to make it clearer than it sometimes is in this literature whether we're speaking just as ordinary humans with intuitions or specifically as moral philosophers, with special reasons for emphasizing moral responsibility.

[hilzoy]

Kip -- first, a clarification: I am not actually hostile to the project of finding out what people mean by freedom (or some other contested concept.) There are points in an argument where that seems to me exactly the right thing to do. In the case of free will, I suspect that what such an investigation will show is that the ordinary concept of freedom has inconsistent strains in it (or rather: strains that are inconsistent in this world, though perhaps jointly realizable in some other.) I am hostile to the idea that that means that we should just decide that it has no application, but that's a different point.

Second: About 'different' concepts and freedom-sub-Hilary and the like: I suppose I might be working with a different idea of 'the same concept' than you. At any rate, I agree with Paul Torek above:

"A concept can survive the discovery that some of the supposed features turn out not to exist. Dennett, if I recall, discusses the survival of 'mass' through the Einsteinian revolution, in Freedom Evolves. The question whether we've discovered that mass is different from what we had thought, or whether we've discovered that mass does not exist, amounts to whether the best candidate for 'mass' in an Einsteinian universe has enough of what was important about 'mass' to qualify."

The concept of mass in an Einsteinian universe would not be exactly identical to the earlier concept of mass. If using 'the same concept' requires using an exactly identical concept, then it's not the same concept, and Einstein's theories imply that there is no mass. If, on the other hand, it's something more like 'that concept that, of all the versions on offer, is the best candidate for mass in an Einsteinian universe', then we can go on talking about mass.

I have never been talking about replacing the concept of free will with the concept of, I don't know, A-1 steak sauce. I take it that the concept of free will involves both compatibilist and incompatibilist elements -- that it means something like: being able to determine one's own conduct, as opposed to having it determined by something else, where the problem is the implicit assumption that when I determine my own conduct, nothing else is determining it. If what matters here is the self-determination part, then compatibilism can work. If what's essential about it is the part about independence from outside determination, then it can't. Either interpretation would preserve something central to the concept of freedom while sacrificing something else. It's just a question of which part to preserve. And that, I take it, depends on what we want the concept of freedom for.

Paul: you write this: "I just have this quibble about "it might turn out that free will doesn't matter": I don't think that 'possibility' should be taken very seriously."

I do, actually. Or, at least: I think that if we try to spell out the coherent positive accounts of freedom of the will that underlie different 'continuations' of our ordinary concept, one will presumably be libertarian, and I am not convinced that those features of a libertarian account that go beyond a compatibilist account will in fact be connected to anything morally interesting, once we see what this sort of freedom would actually be like. (Familiar worries about why, for instance, Robert Kane's various indeterministic thingos should actually matter, morally.) In many cases, I doubt that the worked-out versions would even be consistent with what they're supposed to secure.

Otherwise, though, you made my post a lot clearer than I did ;) Thanks.

[Patricia Greenspan]

Hilary -- I just have to note that in your last few paragraphs just above you immediately change "doesn't matter" to "doesn't matter morally." Since when is morally the only way of mattering?

[hilzoy]

Patricia: an artifact of two things: (a) not being able to proofread well in TypePad's comments, and (b) (last 3 paras.) responding to Paul's point, which is largely (though not entirely) about things that matter morally. I don't, in fact, think libertarianism adds much that matters, period; and I think it probably subtracts -- at least in those versions that can be spelled out in detail, unlike some of the less worked out versions of agent causation libertarianism.

[Patricia Greenspan]

Hilary -- I understand about the proofreading; I've mainly given up on that.

Perhaps all extant versions of libertarianism *and* compatibilism subtract *something* we want from our ordinary notion of freedom -- which, as noted, might contain inconsistencies. But whether or not it's readily spelled out in detail, something like Nozick's notion of "originative value" (see pp. 311ff. in *Philosophical Explanations*; I like his painter example on pp. 312f.) seems to capture something that matters to us, or to many of us, as individuals, even if it's of no significance to moral responsibility. (I should note that I don't find Nozick's defense of libertarianism at all adequate.) We can live without it -- it doesn't seem to figure in what's thought of as an eastern view of things (and incidentally, it would be interesting to see some cross-cultural data on free will intuitions) -- but if we had to live without it for the sake of consistency with the notion of moral responsibility that you want, many would experience this as a loss. Maybe part of what's at issue here is a conflict between 1st- and 3rd-person standpoints of valuation.

[Paul Torek]

Patricia, to clarify:

I am affirming the strong intuitiveness of "free will is worth wanting" while allowing that "moral responsibility is worth wanting" is much less intuitive. I want to concede Pereboom's point (quoted by Kip above) that "It could be, for instance, that we are free in the sense required for moral responsibility, but since being free in this sense is not especially valuable to us, we would not want it much." I suppose the extent to which people desire to be morally responsible depends largely on whether they think their deeds have been good ones ;)

Is a libertarian account of "originative value" the best one? I'm not sure a painter who does something wildly different from anyone who'd gone before, creates something less valuable than one who creates an idea or method literally ex nihilo.

Hilary,

I agree that many well-developed libertarian theories give us something of little moral interest, or even something that seems to merit no significant place in our life's concerns. And yes, one can see how these theories develop from conceptual intuitions that we do tend to have. But in my view, that's precisely the time to wield Dennett's club (free will is worth wanting). If one is sacrificing the near and dear intuition that free will is worth wanting, one had better be able to argue that this is necessary in order to save other intuitions which are collectively stronger.

[Patricia Greenspan]

Paul -- I think there's an interesting shift in what you just wrote between first- and third-person standpoints of assessment, of the sort I meant to allude to.

Being free in the sense required of moral responsibility may not be something people want *for themselves* (unless, as you say, they think their deeds have been good ones), but they may still want and need it to apply to others.

By contrast, in valuing highly the painter who does something different from anything that had gone before (but is acting just as a conduit of past causes, as I think you mean to assume), we're talking about a third party. I'd of course be glad there was such a person. But if I were in the painter's position, I might place somewhat *less* value on my creative act (not particularly *myself* -- incidentally, how do others manage to type italics on the blog?) than I would if I saw the act as insufficiently explained by prior influences, and in that sense a "new beginning," originating in me. Note that this last possibility is distinct from creation *ex nihilo*: there can and should be influences on a creative act that one might cite as causes; they just would not be sufficient, or determining, causes.

I don't mean to deny, as I say, that one can get rid of this concern. Perhaps it would even be healthier just to view oneself as a part of the causal flow. But I guess I wouldn't give up on the alternative simply because it's hard to articulate clearly the reasons why it should matter -- to oneself, if not to onlookers.

[Kip Werking]

Hilary,

If you agree that it is important and interesting to see what the folk actually mean by “free will”, and see if philosophers are using that term consistent with its ordinary usage (if any), as you seem to agree, then our original dispute/difference is resolved.

But let’s move on to the newer issue: the revisionist challenge. It is remarkable that Manuel Vargas hasn’t posted in this thread, because it seems to discuss the revisionist themes in his work so clearly. According to the revisionist challenge, philosophers may want to use “free will” in a way that is inconsistent, at least somewhat inconsistent, with its ordinary usage, because this use is still proper. By analogy, Dennett cites “mass”, and Vargas cites “whales.”

We used to think whales were fish. But then we discovered that they are not fish; they’re mammals. But, as Vargas points out, this doesn’t mean whales don’t exist. They do exist. They’re just not quite what we thought they were. Dennett makes the same point about “mass”, but I think the analogy is more strained there.

One temptation is to try and poke holes in those analogies. But this is probably an imprudent strategy. Even if the whale and mass examples have problems, there are probably some such examples of properly evolving concepts.

A better response, I think, is to note the revisionist hypothesis is just that, a hypothesis, and the revisionist needs to more to show that “free will” is like mass and whales in this way. Certainly, it would solve the compatibilist’s problems, and it presents a tantalizing possibility. But is the possibility actualized? The problem, of course, is that one can immediately think of other examples, and other analogies, according to which the concept’s potential evolution is obviously *improper*.

This is what I have called the unicorn danger. Perhaps free will is not like mass or whales. Perhaps it is like unicorns. And instead of free will properly evolving like the concept mass or whale, certain philosophers are trying to improperly evolve it, like the term “unicorn” referring to horses.

You note that the difference between the folk’s somewhat incoherent use of “free will” and your own suggested use is not so great: you are not talking about “A-1 steak sauce.” But please also note that the difference between many concepts is not so great: unicorns and horses, apples and oranges, CDs and DVDs and so on. I’m sure even more tiny and subtle, but essential, differences exists between many things we would both agree are distinct concepts.

So, at best, I think the revisionist has posed a tantalizing possibility: free will might be like mass or whales. But there is more work to do, because free will might also be like unicorns. And how do we decide between them? This is honestly a question I asked Manuel Vargas in an email about three months ago.

One can note at least some one potential difference between free will and mass/whales. Mass and whales both refer to things that are findable and undeniably exist in the world. So our attitude towards these things is something like “mass and whales are these things that I can find at X, Y, and/or Z, at times A, B, and/or C. Now, I tend to think that mass and whales have properties 1, 2, 3, 4… but my belief in all of these attributes is fallible. After all, I might go inspect mass/whales at X, during A, and discover that, to my surprise, they don’t actually have property 2, but they do have property 22, which I had not known about. My beliefs about mass/whales existing and being open to inspection in certain ways are governing here, and any beliefs about what these things actually are is subordinate to whatever an inspection of those things might show. Thus, inspecting the empirical world is the ultimate arbiter of what mass or a whale is.”

But, the anti-revisionist can respond, it is remarkable how the concept “free will” is *not* like this. There is no thing “free will”, which all parties agree undeniably exists and can reliably locate and inspect in the empirical world, such that we subordinate any intuition we have about “free will” to the results of going out into the world, finding free will, and inspecting it. If there is any controversy about free will, we can’t say “well, you raise an interesting question, anti-revisionist, I am pretty confident that free will entails TNR, but I could be wrong; I do know, however, that free will always hangs out at the pier around midnight, so tonight I’ll go to the pier, inspect free will, and actually find out if free will does entail TNR.” We can’t do that. We wouldn’t know where to go, we wouldn’t know what to inspect, we don’t even agree that there is such a thing free will that we could inspect.

Instead, free will seems all the more like *unicorns*. Some people think they exist (hopefully not too many, but the same should be true of *God*, which many people do think exists); others hotly contest this claim. If there is any controversy about what unicorns actually are, or what God actually is, we can’t go inspect the world and revise our fallible beliefs about them. No, the only way (to my knowledge) to decide what unicorns or God actually means is to ask people: to determine what the common usage of those terms is. The revisionist trick, or escape route, whereby beliefs about something can be revised in accordance with the higher-level belief that “X exists and can reliably be located and inspected, and such inspections are the ultimate authority on what X is”, does not seem to be available here.

That is one potentially fatal response to the revisionist challenge. There may be others. But I certainly don’t think the revisionist has succeeded in establishing revisionism about “free will”. The unicorn danger is still present.

Now, I do want to address an alternative line of argument given by revisionists like Hilary and Manuel: the functional role of a concept. According to these philosopher, roughly speaking, the meaning of “free will” should be determined according to what role we want it to play in our lives.

I am not at all convinced that this functional analysis of concepts, according to the roles we want them to play, is strong enough to be consistent with standard language practice. It seems to improperly infect questions of what does and does exist, with questions of what we wish was so and wish was not so. Those two questions, of course, don’t have anything to do with each other, and should be kept separated (if only to avoid the danger of wishful thinking).

But that is a question for another day. Let’s put that worry aside for a moment. My point is just that: even if we grant the revisionist this functional analysis of concepts, I don’t think, and all or most libertarians and eliminativists don’t think, that your weak concepts can fulfill the role that we want “free will” to play in our lives. This is what Al Mele’s Zygote Argument (and arguments like it going back to Bramhall) shows so well. You don’t seem to just be cutting the fat off the concept; you’re cutting something essential. You can note how stronger concepts of free will seem non-sensical or logically incoherent (and ultimately they are), but one aspect of these concepts is crystal clear, if only by stipulation: an agent with this kind of free will could never be a puppet, molded and designed by a puppet master, like Ernie is in the Zygote Argument.

Nor do I think these fantastic scenarios are so idiosyncratic and not to be taken too seriously; I think they are symptomatic of serious underlying deficiencies in what compatibilists regard “free will” as meaning. Whatever stronger definitions of free will might otherwise mean, or not mean, it was understood that an agent with such free will could *not* be the puppet of some cosmic puppetmaster, like the agent in the Zygote Argument. This is *also* what people wanted and expected from free will, and why the argument rattles even smart philosophers who are sympathetic to compatibilism, like Al Mele, and shakes their faith in the viability of that view, including its revisionist varieties. So, it seems to me, even granting the functional analysis of concepts that the revisionist would like to use, weaker compatibilist notions of “free will” do not pass muster, and do not do justice to all that we wanted or expected free will to do in our lives.

In summary, there remains the concern, discussed earlier, that there is a disanalogy between free will and the other evolving concepts that revisionists would like to compare free will to: mass, whales, etc. The unicorn danger is still present and that worry is perhaps fatal to the revisionist’s project. And secondly, even if we grant the revisionist the functional analysis of concepts, their weaker notions of free will do not seem to do all of the work we wanted and expected free will to do in our lives.

[Kip Werking]

Paul,

Your position is based on the intuition that free will is worth wanting—an intuition that I feel is worthless. As I mentioned earlier (point 3 in my long Dec 18 post), desirability is not a predicate, in the same way that existence is not a predicate. Now, I apply what I call the fundamental distinction, between concepts and inferences, to this fact about desirability not being a predicate: concepts are determined by common usage, and are popularity contests in this sense, but inferences are not determined by common usage, they are determined by logic, and even if 99% of people think some inference is correct, they might still all be wrong. In other words, the same fundamental distinction that lets me say “to determine what free will is, we need to ask the folk” *also* lets me say “if everyone thinks free will is desirable, well they just might all be wrong, just as people were all wrong to believe that the sun moved around the earth and that humans did not evolve from other species.”

So, if the folk say (and I am not sure they would): free will is, in part, the power to make decisions entirely free of external influence, then that common usage would govern. But, if the folk all say: free will, that’s some fashionable stuff, I gotta get me some of that, then the popularity of free will *doesn’t* govern. Compare:

A. Everyone says “dirt” is wonderful and means “hugs and kisses.” I want some dirt!
B. Everyone says “dirt” is wonderful and means dirt. I want some dirt!

In A, dirt actually means “hugs and kisses”—no matter what any faction of philosophers say—because common usage determines what “dirt” means. Likewise, in B, dirt actually means dirt. But, in B, when everyone says dirt is wonderful, common belief does *not* have the same authority. In B, we don’t thereby say “wow, everyone thinks dirt is wonderful, therefore it must actually be wonderful.” And we certainly don’t inspect real dirt, find that it isn’t wonderful, and conclude “wait! This isn’t wonderful! People must have been talking about something else!” No, we just say: everyone was wrong. I don’t have anything special against dirt, but it seems largely boring and worthless. If everyone thinks dirt is wonderful, I’m not going to revise my beliefs about dirt, I’m just going to say they are wrong.

Now, you strangely acknowledge this point with respect to *moral responsibility*, but resist it with respect to for free will. This reminds me of Fischer’s distinction between “morally responsible” and “blameworthy for a morally-non-neutral act”—such distinctions have never impressed me much. Even if I granted you that desirability can be determined by popularity contests (contra my argument above), I doubt that the folk have any stronger belief about the desirability of moral responsibility versus the desirability of free will. But I would love to be proved wrong. In any case, I don’t grant that desirability can be determined by popularity contests in this way, and I think my argument above is cogent and makes clear that there can be no special treatment for “free will”, as opposed to moral responsibility: desirability, with respect to any given thing, is not a predicate, just as existence is not a predictate, and cannot be determined according to common usage or a popularity contest or some such thing.

[Paul Torek]

Kip, I think that if everyone around you says that dirt is wonderful, then charitable interpretation obliges you to consider seriously the hypothesis that "dirt" around there doesn't mean dirt. Or that they're all farmers, and dirt really is vital.

This would probably be a good place to register my view that the distinction between concepts and inferences is something like that between two sides of the same coin. Consider the appeal of conceptual role semantics, described at this website:
http://www.iep.utm.edu/c/conc-rol.htm

I'm not saying that conceptual role semantics is the revealed truth, but if it's even remotely close, I doubt that we can draw sharp bright lines using the distinction between concepts and inferences.

Consider for example the inference from "A and B are holding hands" to "A and B are touching". I know of no formal logical system which licenses this inference, yet it seems valid. If it is valid, it is precisely the concepts of "holding hands" and "touching" that make it so. Now suppose that there is some controversy about the concepts, or the inference, or both. If we try to ascertain what the folk commonly mean by these phrases, aren't their inferences (or refusals to infer) part of the data by which we so ascertain?

Back to desirability: I just don't get it why it matters whether desirability is a predicate. Nobody is trying to run an Ontological Argument here. Relativize "desirability" all you want, if that suits you; I don't see how it makes a difference.

[Kip Werking]

Paul,

When you write:

"[I]f everyone around you says that dirt is wonderful, then charitable interpretation obliges you to consider seriously the hypothesis that "dirt" around there doesn't mean dirt. Or that they're all farmers, and dirt really is vital."

You are just resisting the thought experiment, which stipulates that "dirt" does mean dirt. Certainly this world might different than our own world. But it is different because the people in our world don't seem to have mistaken beliefs about dirt, and not because the people in our world don't have mistaken beliefs about anything.

Again, the folk belief all sorts of absurd things. Instead of writing "dirt is wonderful", I could have written "the sun moves around the earth" or "people are the special creations of the Christian God, and did not evolve from other species." Most people believed those, but to say "charitable interpretation obliges you to consider seriously the hypothesis that 'sun' around there doesn't mean sun" is the wrong response.

However, I can grant that most people believing something is *some* evidence of that proposition being true. It does convey some information. I'll come back to this.

You want to know why the claim "desirability is not a predicate" is not relevant here. I'll try to explain one last time, and if you still don't agree, then maybe we can talk about it again some other time.

The starting point is this fundamental truth about language: the meaning of terms is in accordance with their common usage. This leads to the conclusion: the question "most people use some term to mean X, are the folk right to do this" is a *meaningless* question. Just the fact that most people use the term that way *proves* that that is what the term means. In other words, we are starting off with *certainty*. We know the answers to such questions, just by seeing what most people do. We don't have to ask the further question "but are most people right to do this?" That is a meaningless question.

Once we get beyond predicates, like "squares have four corners", and move into further questions/inferences, like "squares exist" or "squares are desirable", then we have left this realm of certainty. *That* is why the difference is relevant. If someone wants to argue that squares exist, or are desirable, s/he can no longer point to how people use the term "square". S/he will have to go above and beyond, and confront the possibility that, even if all or most people agree that squares exist or are desirable, they are all just wrong. We might call this the Copernican or Darwinian danger: the worry that, even though a belief is popular, it is mistaken. In contrast, predicates, as defined by common usage, are *invulnerable* to the Copernican/Darwinian danger. There is no further question to ask "well, sure, everybody believes squares have four corners... but are they right." Common usage is always right, by fiat!

That is why this distinction between concepts/inferences is relevant. For example, if compatibilists want to make claims about what free will is (predicates), then they are *bound* by the term's common usage (if any). Similarly, if compatibilists want to make claims about whether free will exists, or is desirable, then they must rely upon more, to support such claims, than the fact that most people believe free will exists or is desirable. But, in this thread, it seems to me that you and other compatibilists would like to do the opposite: you would argue that free will means something different than what most people say it means, and you would support the claim "free will" is desirable with the fact that most people think it is desirable. (You may have other arguments for thinking that free will is desirable, but I have not seen them, and any such arguments will need to anticipate the possibility that free will actually means, or tends to mean, something logically impossible.) This seems to get things exactly backwards.

Now, getting back to the popularity of a belief being some evidence that it is true: I can grant that the popularity of a belief is some evidence that it is true. Certainly, the number of popular true claims seems to vastly outnumber the number of popular false claims. It seems to be a useful heuristic. But it is just a heuristic, and it has its limits. The Copernican/Darwinian danger looms: maybe everyone is just wrong to think that free will is desirable. And so, just the fact that this belief is popular, without *more*, is not sufficient to demonstrate its truth (contrast this with the popular belief that squares have four corners, which is sufficient to demonstrate its truth). Indeed, eliminativists like Pereboom and myself think that the Copernican/Darwinian danger is actualized in this case. Once we have confronted the proponent of free will's desirability, and said "wait, what about the Copernican/Darwinian danger, the popularity of this belief is evidence, but it is not definitive", and put the compatibilist to hir proof, then it only seems reasonable that the compatibilist would say "free will is desirable, and not just because most people think it is, but because of reasons X, Y, and Z." But compatibilists have not done this, at least in this thread (Fischer makes the interesting argument that acting freely is valuable because it allows one to “express a sentence” in “the book of [one’s] life.”)

That is why the distinction between concepts and inferences is relevant.

Finally, let me add:

1. I'm not a philosopher of language. I don't know how conceptual role semantics, if true, would affect my arguments above. The article you linked to is quite long and technical, but intriguing, and I haven't had time to absord it yet.
2. The "holding hands" seems to be something like a borderline case. Perhaps "touching" could be analyzed either way, as part of the concept, or as an inference, albeit an inference that most people are extremely likely to correctly make. However, even if things are not clear in that case, I don't think I need to draw such fine lines. Things nevertheless seem clear in our case (what free will means versus whether it is desirable), and my arguments above should show that, and the notion of the Copernican/Darwinian danger show why the distinction is important.

[Paul Torek]

Kip,

I'm no philosopher of language either. Philosophy of language usually makes my brain hurt. But I figure since you're wandering into that territory, and describing things that may contradict some of the maps I've read, I should complain about the apparent conflict, lest I be led astray.

I think "free will" may be a lot like "holding hands" in terms of borderline conceptual entailments. Only, the greater complexity of usage of "free will" allows, in principle at least, for multiple borderline-entailments, perhaps with varying degrees along an analytic/synthetic spectrum.

On desirability: virtue is desirable. Goodness is desirable. Safety is desirable. Bliss is desirable. Autonomy is desirable. Not all of these are inferences. I think it would be closer to the truth to say none of them are. It would be closer still to say that all of them are explications of part of the conceptual content of the subject term, yet are also more than that.

So it's not true that "desirable" can't be any part of the meaning of any terms. But more importantly, it's not necessary for desirability to be part of the meaning of "X" in order for "X is desirable" to be a strong desideratum for any theory of X.

Even if there are some slight room for skepticism about whether autonomy is desirable, or a few people who can with perfect rationality dismiss it, it's still extremely probably desirable for most of us. And any account of free will which disconnects it from autonomy is in serious trouble with our linguistic intuitions. That's enough to embed "free will is worth wanting" into a large chunk of the usage of the term "free will". That doesn't make "free will is worth wanting" a fixed point, but it does mean that the payoff for rejecting it needs to be very large. We can't determine the meaning of "free will" while turning a blind eye to this portion of the usage.

[Kip Werking]

Paul,

I don't want to beat a dead horse, but here are some responses to your concerns. You pose a stronger challenge than I anticipated, to what I thought was a pretty obvious point.

1. As I said, I don't think "free will" is like "holding hands". I'm not sure how to resolve this dispute.

2. About your examples of things that are desirable, by definition, first (and I repeat this, because it is important), even if I grant that these cases pose a problem for the fundamental distinction, I don't have to grant that they are relevantly similar to free will. It may be that, whatever other murky cases there are, free will is a pretty clear cut one. I certainly think so. The fact that we seem to ultimately disagree about this just lends further support to Richard Double's (former?) view that free will has no consistent, coherent meaning.

3. It is not true that virtue, safety, bliss, and autonomy are desirable by definition. They are desirable, if they are desirable, because we are normal human beings who value these things. I can *easily* imagine psychopaths or other strange creatures that disvalue, or even abhor, virtue, safety, and autonomy. So there is an inferential step that must take place before you conclude that these are desirable (if you do). You cannot find "desirable" within the four corners of the document giving their definitions. What if the document only exists in a world with psychopaths in it, and they all hate virtue? And this inferential step introduces a vulnerability, where people can make mistaken inference (again, contrast this with "the meaning of a square is, among other things, something with four corners"; there is no inferential step and no vulnerability).

I have difficulty conceiving of a creature that would not desire bliss, but it is not clear to me that this represents an interesting truth about the space of all possible minds, or simply a failure of my imagination. Perhaps it just represents my utilitarian leanings. The mind can be much more stange and wonderful than I can conceive (e.g. I've read of mental patients who can report the experience of pain without any aversive content: they know they are feeling pain, and can identify it, but they don't care).

But I am not sure that "desirable" means "would desire x" so much as it means "*should* desire X". And in that case, I think one can take the same skeptical attitude one takes towards virtue, autonomy, etc. It is interesting to note that Nietzsche, for one, was not a utilitarian, did not value bliss, and emphasized the importance of suffering in making great individuals. That, alone, may prove a counterexample to your point, but I'm not sure my position here is perfectly sound.

4. You wrote:

"But more importantly, it's not necessary for desirability to be part of the meaning of "X" in order for "X is desirable" to be a strong desideratum for any theory of X."

I'm not sure what you mean by "to be a strong desideratum for any theory of X." But I'm not sure I am denying that desirability should automatically not be part of a theory of free will. I'm saying: "if you want to say that free will is desirable, then prove it, and your proof better be something more, or better, than saying most people think it is desirable."

5. I find autonomy to be such a philosophical term of art, without any coherent or consistent meaning, that it is almost worthless to talk about as a single thing. Vargas has raised similar concerns, reviewing a recent book about autonomy.

6. It occurs to me that you might come to agree with me more, if you would, if you better understood my own personal hunch: free will has compatibilist strains, and incompatibilist strains (or weaker and stronger elements), as Hilary Bok said. It is not devoid of content. But something about it, some incompatibilist element, is *gumming up the works*, and putting a wrench in the machinery, such that it cannot actually exist. This may be a transfer principle like TNR or Beta. Now suppose that most people go about their lives, with a *vague* understanding of what free will is, without *absorbing* how the stronger element gums up the works, so to speak. First of all, if they say "free will is desirable", they say it after making an inference, and not because free will is desirable by definition. I can easily imagine psychopaths or strange creatures that don't desire free will, and so it isn't desirable by definition, but it's desirability is contingent upon certain facts about human psychology and so on. Secondly, given that they fail to absorb how free will is impossible, because certain stronger/incompatibilist aspects of it gum up the works, free will is actually not desirable (in the sense that all logical contradictions and impossibilities are undesirable), and so, when people made this inference, from what free will is to whether it was desirable, they made a mistake. This is my own suspicion, and it may be Pereboom's own view, as indicated by his introduction in LWFW.

Note that it is easy to see why people would make this mistake: people want more control and unpredictability and responsibility, and they are greedy, and they will keep asking for it, and believing it, past the point where it no longer makes any sense. We are talking about beliefs that intimately relate to human nature, human emotions, human aspirations, our place in the universe, and life hopes. We can probably agree that when people die, they're dead, and there is no afterlife. Yet most people actually believe in an afterlife (or say they do). On the view I have sketched here, people make a similar mistake about the desirability of free will as they do about believing in an afterlife: their uncritical inspection of vague ideas provides them with emotional comfort, and they let these emotions bias their reasoning process.

Now, everything I have just sketched is a hypothesis. But it is a hypothesis that your view does not seem to leave room for. Isn't that a problem?

For example, I can imagine you and Hilary coming back and saying: "Well, if almost everyone believes free will requires TNR, and if almost everyone believes that free will is desirable, then since the desirability aspect is so important, these people are just mistaken to think that free will requires TNR, and so I am going to conveniently remove that element. We'll peform a little linguistic surgery, and now we get our free will, and our desirability too."

In contrast, I would say: "Desirability, despite what any examples about virtue and autonomy try to prove, is not a predicate, and so beliefs about it are more suspect than beliefs about what free will actually is. So that if people passionately believe that free will requires TNR and they passionately believe that free will is desirable, TNR trumps desirability. Thus free will isn't actually desirable, people were wrong to think so, although it obvious why they made that mistake, and also free will doesn't actually exist."

[Caveat: I can hear Eddy reminding me now that the evidence in favor of TNR being popular is mixed at best.]

I suppose whether you come to agree with me will depend upon how viable you think your hand-holding, autonomy, virtue, etc. counter-examples are. I've raised my objections to those, and explained my view, so there may not be much more for me (or us) to say. I have really enjoyed the dialogue though.

[Paul Torek]

Kip,

I think in some sense part of what we are doing is rehashing the analytic/synthetic debate. Part of the reason I can't accept some of your arguments and you can't accept some of mine, is that I think "analytic or conceptual truth" versus "inference" is a matter of degree, and you think it's a hard and fast distinction. So if I say that "desirable" is part of the meaning of "virtue", you take that as implying a hard and fast rule, such that virtue without desirability would have to be like a bicycle designed to have no wheels. Whereas, I take that as implying only that if you strip "desirability" out of "virtue", you get a mangled version of the latter concept - what some philosophers call "the inverted commas sense of 'virtue'". Is inverted-commas 'virtue' still virtue? I feel no obligation to give that question a flat out yes or no answer. I prefer to say "not exactly".

It's not that I fully agree with Quine and deny that there are any plainly-and-simply-analytic truths. It's just that you won't find them in most of philosophy, outside of logic and metalogic. No analysis of terms like "free will" will ever neatly and tidily capture their exact conceptual content, I'm afraid, no matter how carefully lexicographers and psychologist examine the usage of these terms. The best, if not the only thing we can do, is follow Hilary's advice, and go constructive.

On the hypothesis that "free will" picks out a logically impossible combination of conditions, your hypothesis of how people manage to ignore or evade this fact is perfectly plausible. People can fail to recognize contradictions, especially if the logic is complex and the motivation for denial/rationalization is strong.

On the other hand, people can also call "contradictory" that which is not, and this sort of error is much more common. Compare deductive inference, which is valid IFF the conjuction of premises plus the denial of the conclusion is inconsistent. There is a long list of fallacies - argument types which many people regard as valid, but which are not. Thus, there are many sets of statements which would be regarded as inconsistent but which are really not. In contrast, there is a short list of "trulacies": valid argument types which are commonly seen as invalid. Going by base rates alone, then, whenever you see a debate between some people who think a set of statements is inconsistent and others who think it is not, your money should be on the latter group.

Not that we should go by base rates alone.

I do agree that there are "compatibilist" and "incompatibilist" strains or families of intuitions. If somehow both could be beautifully accomodated in one coherent account, that would be ideal. I don't think that's possible however, so each camp ought to try to save as many of the "phenomena" as possible. If you can't endorse an intuition, at least try to explain it as a non-idiotic mistaken inference from some true belief, or as a sane if flawed interpretation of experience.

Only if all such attempts from either side fail miserably and monotonously, and only if neither lost set of intuitions can be jettisoned without major disruption, should we begin to suspect that there is no one coherent concept of free will. It's too early for such a verdict.

[Bart van Beek]

I'm thoroughly enjoying the discussion here as well and am in fact using it as an inspiration for my exam on free will. Being usually more involved with philosophy of language, I tend to agree with Kip's line of argument and I believe his Nietzsche example in fact lends precisely the sort of support for his position that is needed.

First let me say however that I disagree with Kip on one particular point and that is the question whether or not `to be desirable' is a predicate? Obviously as it seems to take a noun as its argument, it intuitively appears to be a predicate. (Notice I have shifted here from `desirability' to `to be desirable', as there is no need to discuss predicatehood of the former, it is obviously not a predicates.) So Kip must have his reasons for denying it predicatehood. It seems that his only reason is an assumed analogy between the predicates `to be desirable' and `to exist'. I do not see this analogy however, because it would appear to me that the reason to deny predicatehood to `to exist' is not present in the case of `to be desirable'. This reason is the well know problem of Plato's beard, which pops up in the context of negative existential statements. The problem arises because something must in a certain sense exist before we can assert anything of it, but if it does not exist in any way, how can we then say of it that it does not exist. None of these problems arise in the case of `to be desirable'.

It is quite possible however that this is completely irrelevant because I take (part of)the discussion to be about the question if desirability is included in the meaning of `free will'. Assume that this is the case. Then, as an analogy, consider Nietzche, a philosopher who supposedly studied the meaning of `good' and `evil' and their historical developments with a great level of attention. Someone with a significant amount of knowledge of the contexts in which these words were being used. This philosopher then states `good is not desirable´ (I'm not sure he actually did this, but we can assume he did so nonetheless). What would we have to say to Nietzsche if desirability was part of the meaning of `good'? I think it would be something like this: ``Dear Mr Nietzsche, you are saying of something that is desirable that is is not-desirable, that is the sort of manifest contradiction that even you cannot allow, either you don't understand what `good' means or your criticism is inconsistent.'' We know that Nietzsche knows what `good' means, so we would have to come to the conclusion that Nietzsche is not giving a carefully weighed and trenchant criticism of a persistent idea, but is in fact giving an inconsistent argument. I don't think that this is the way most people would view Nietzsche's criticism at all and I think it shouldn't be viewed like this either. Nietzsche knows perfectly well what `good' means, he is not disputing its semantics, nor is his argument inconsistent, he just doesn't think good is desirable.

So we have to ask ourselves, does `undesirable free-will' really constitute a condtradictio in adiecto and do people who claim that free will is undesirable not understand what everybody else means by free will, as the above argument would seem to suggest is the only possibility if we found there arguments otherwise cogent, or does the notion of `undesirable free will' merely make us feel a bit strange. I think the Nietzsche example is good evidence that `undesirable free will' is a strange but not an inconsistent notion. Furthermore I think that if we realize that desirability is not part of the meaning of `free will', we can easily explain this fact. For only under the assumption that it is not part of the meaning of `free will', does it make sense to speak of undesirable free will, but if free will really is undesirable it is practically useless to most of us. For in most of the contexts in which we (or at least those philosophers who feel free will is desirable) use `free will' we like to use it as something desirable. If it would turn out that this is not possible the possible contexts in which we could consistently use free will would be seriously limited and hence the concept would be mangled. This does not however make someone who claims that free will is undesirable wrong about its semantics.

[RICHARD DOUBLE]

I wish to reconcile these claims:

A] The intuitions of nonphilosophers regarding, e.g., the acceptability of compatibilism, given their understanding of the meaning of moral responsibility, do not matter with respect to which philosophical views one should adopt.
B] The intuitions of philosophical experts in these topics do matter [and their conflicts are important data in favor of subjectivism of moral responsibility and desert-enabling free will].
C] The meanings we assign to words are conventional in the sense that they are not to be graded relative to some scientific or supernatural essentialist criterion of correctness.

How can one disparage the importance of common speech [A] and applaud the insights of the cogniscenti [B] without endorsing a Kripkean, Platonic, or some other inflated view of language [C]?

Consider Putnam’s thought experiment in which all the entities we call ‘cats’ are remote-controlled artifacts, and suppose we rely on our intuitions about cats to determine whether to count the artifacts as cats or not. Here are 4 views: [1] our intuitions count heavily-to-decisively and it matters which we decide. 2] Our intuitions do not count heavily-to-decisively and it matters which we decide. [3] Our intuitions count heavily-to-decisively and it does not matter which we decide. [4] Our intuitions do not count heavily-to-decisively and it does not matter which we decide.

Regarding cats, water, and other descriptive terms, I believe it does not matter philosophically which of these four positions we take [whether we take our intuitions as important and whether we use a microstructural criterion]. As long as we get straight about the nonlinguistic facts, i.e., that our pets are nonbiological, it does not matter whether we say they are cats or not, and it doesn’t matter whether we decide this on ‘phenomenological,’ ‘functional,’ or ‘microstructural’ grounds. Only [1] denies my claim; but [1] seems committed to a creepy version of the doctrine that there are deeply right meanings of descriptive words.
Compare cats and water with examples of moral terms such as moral obligation and moral responsibility. In the latter case it matters which of the four positions we take. I believe [1] is the front runner, because [i] here we care about calling things ‘by their right names’ and [ii] intuitions are vital when we make moral judgments. In the morality case, [1] avoids the ‘right meaning of words objection’ by giving a relevant way of determining them: moral intuitions are important because morality is based on moral intuitions, attitudes, and feelings—something not true of felines.
One may find circularity in my argument because it assumes the fact-value distinction. I plead partially guilty; but I believe my argument supports the distinction more than the distinction supports it. [Educated] intuitions count in moral questions, because our aim is to clarify moral thought about desert, blame, etc., which are matters about which have deep feelings that are revealed by our moral intuitions. We care so much more that, e.g., say to those who disagree with us that are promulgating wretched subterfuge or engaging in incoherence. So we discard the senses of moral terms that do not cut to the quick. The price one pays for such strong convictions, though, is that these heartfelt feelings are better seen as aimed at subjective concepts. Educated intuitions contribute only to taxonical neatness when dealing with concepts for objective [non-psychological] entities.

[Paul Torek]

Bart,

Is there room on your view for family resemblance concepts? If so, Consider the classic example "game". Would it be correct to say, for each of the qualities such as playfulness, or competition, or so on that help us to recognize something as a game, that "playfulness is no part of the meaning of 'game'"?

[Kip Werking]

I want to respond to the three very thoughtful posts from Paul Torek, Bart van Beek, and Richard Double.

For Paul Torek:

You wrote: "I do agree that there are "compatibilist" and "incompatibilist" strains or families of intuitions. If somehow both could be beautifully accomodated in one coherent account, that would be ideal. I don't think that's possible however, so each camp ought to try to save as many of the "phenomena" as possible."

First, this presents a sort of false dichotomy: an account of free will can have the compatibilist elements or the incompatibilist elements, but not both. But incompatibilist accounts of free will *can* have both elements and, in fact, *do*: incompatibilists are famously criticized for taking the working features of compatibilism and just adding indeterminacy to it, which is just distracting and worthless (see, e.g. Watson's critique of Kane in his great Soft Libertarianism and Hard Compatibilism). And in this thread (and elsewhere) I have not advocated the use of incompatibilist or stronger notions of free will (if that is, in fact, how most people use the term) *at the expense of the elements in weaker concepts*, but rather in addition to them. You seem to posing this issue as a choice between:

Compatibilist conception of free will: A, B, C
Incompatibilist conception of free will: D, E, F

But I tend to think of it as:

Compatibilist conception of free will: A, B, C
Incompatibilist conception of free will: A, B, C, *D*

Where, again, D is some extra element, probably a transfer principle or historical consideration (like TNR or Beta) that *gums up the works* of the rest of the concept.

Ultimately, you say "I don't think [it's] possible for both [compatibilist and incompatibilist strains] to be beautifully accommodated in one coherent account." But what you seem to mean by this is: "I don't think [it's] possible for both [compatibilist and incompatibilist strains] to be beautifully accommodated in one coherent account---such that we still have free will." Both elements can be accommodated, and are accommodated, in the stronger and more demanding concepts of free will that incompatibilists (whether of the libertarian or hard determinist tradition).

For example, consider Al Mele’s view on autonomy, as discussed in this thread:

1. HISTORICAL
2. reliable deliberator
3. ideally self-controlled
4. beliefs conducive to informed deliberation
5. [other(s) I may be forgetting]

As far as I can tell, it is the HISTORICAL property that makes the Zygote Argument work. Without it, you would be left with a good hard compatibilism. But note that these other elements (besides HISTORICAL) are *also* elements of a reasonable compatibilism. So I think I it is a false dichotomy to say incompatibilist or compatibilist elements can only be accounted for at the expense of the other.

At this point, it is important to remember a crucial and subtle point that Galen Strawson makes in the Bounds of Freedom: just because a concept entails something logically impossible does *not* mean that it is meaningless or needs to be revised. For example, 2+2=5 is not meaningless; it presents something that can never be true, but we only know that *because* it makes (at least some) sense. Similarly, stronger accounts of free will might entail something logically impossible, but that doesn't mean that, as you wrote: "each camp ought to try to save as many of the "phenomena" as possible." Nothing needs to be saved! The concept works fine the way it is, but it refers to something that cannot exist. If you want to say that something else exists, call it free will*.

You also wrote:

"If you can't endorse an intuition, at least try to explain it as a non-idiotic mistaken inference from some true belief, or as a sane if flawed interpretation of experience."

And so, in case you missed it the first time, I will just repeat the error theory I offered earlier:

"Note that it is easy to see why people would make this mistake: people want more control and unpredictability and responsibility, and they are greedy, and they will keep asking for it, and believing it, past the point where it no longer makes any sense. We are talking about beliefs that intimately relate to human nature, human emotions, human aspirations, our place in the universe, and life hopes. We can probably agree that when people die, they're dead, and there is no afterlife. Yet most people actually believe in an afterlife (or say they do). On the view I have sketched here, people make a similar mistake about the desirability of free will as they do about believing in an afterlife: their uncritical inspection of vague ideas provides them with emotional comfort, and they let these emotions bias their reasoning process."

For Bart van Beek:

I am glad you sympathize with my arguments. As this thread went on and on, and my comments grew longer, I was worried that I speaking more and more rashly (and making less and less sense).

Let me first say that "desirability is not a predicate" may be too strong. I'm not even sure what a "predicate" is, other than the sense in which Kant says that existence is not a predicate (I haven't pretended to be a philosopher of language). However, even if it is slightly like a predicate, as you seem to think, it still seems sufficiently not-predicate-like (call it not-predicate*, if you like) for my argument to still have some force, as you also agree.

Indeed, I think I can even grant Paul's point that the concept/inference distinction is not black and white, but admits of degrees, and *still* my argument will have some force. Suppose, for example (to speak in rough terms), there is a spectrum mapped onto the unit interval from 0 to 1, where 0 is absolutely part of a concept and 1 is absolutely not part of a concept. Suppose further than TNR is .2 and desirability-of-free-will is .8. Now, even if we grant this move from absolutes to a spectrum (and I am not sure that we should), it still seems that, if we had to choose between TNR and desirability, desirability would have to go. The closer we get to 0, the closer we get to saying "there is no question that common usage is wrong about this", and TNR seems so much closer to 0 (even if it is not absolutely at zero), that we would question free will's desirability well before we would ever question TNR.

Now, when I say "desirability is not a predicate", even as a matter of degree, I only mean this in a sense that seems quite obvious and intuitive: the content of free will (whatever that vague content may be) involves powers, ability, freedoms, unpredictability, and such. The question of whether this is a good thing is a question *above* and *beyond* the question of what it is. To repeat myself: the question of what something is, is the sort of question that common usage determines; the question of whether that something, so determined, is valuable or desirable is a further question---the sort of question that people can make mistakes about.

For Richard Double:

You raise concerns about cats, water, etc. I believe all of these things satisfy the alternative analysis I gave for Hilary earlier in this thread. I am going to reproduce that argument at length:

-----------------
One can note at least some one potential difference between free will and mass/whales. Mass and whales both refer to things that are findable and undeniably exist in the world. So our attitude towards these things is something like “mass and whales are these things that I can find at X, Y, and/or Z, at times A, B, and/or C. Now, I tend to think that mass and whales have properties 1, 2, 3, 4… but my belief in all of these attributes is fallible. After all, I might go inspect mass/whales at X, during A, and discover that, to my surprise, they don’t actually have property 2, but they do have property 22, which I had not known about. My beliefs about mass/whales existing and being open to inspection in certain ways are governing here, and any beliefs about what these things actually are is subordinate to whatever an inspection of those things might show. Thus, inspecting the empirical world is the ultimate arbiter of what mass or a whale is.”

But, the anti-revisionist can respond, it is remarkable how the concept “free will” is *not* like this. There is no thing “free will”, which all parties agree undeniably exists and can reliably locate and inspect in the empirical world, such that we subordinate any intuition we have about “free will” to the results of going out into the world, finding free will, and inspecting it. If there is any controversy about free will, we can’t say “well, you raise an interesting question, anti-revisionist, I am pretty confident that free will entails TNR, but I could be wrong; I do know, however, that free will always hangs out at the pier around midnight, so tonight I’ll go to the pier, inspect free will, and actually find out if free will does entail TNR.” We can’t do that. We wouldn’t know where to go, we wouldn’t know what to inspect, we don’t even agree that there is such a thing free will that we could inspect.

Instead, free will seems all the more like *unicorns*. Some people think they exist (hopefully not too many, but the same should be true of *God*, which many people do think exists); others hotly contest this claim. If there is any controversy about what unicorns actually are, or what God actually is, we can’t go inspect the world and revise our fallible beliefs about them. No, the only way (to my knowledge) to decide what unicorns or God actually means is to ask people: to determine what the common usage of those terms is. The revisionist trick, or escape route, whereby beliefs about something can be revised in accordance with the higher-level belief that “X exists and can reliably be located and inspected, and such inspections are the ultimate authority on what X is”, does not seem to be available here.
-----------------------

Also, I want to emphasize that, when I say "folk intuitions are important", this is not an unqualified statement. I think this is the concern that some (such as you and Hilary) may have about folk intuitions. Suppose I went out and asked a 1000 undergrads, not about come concept, but about some inference (as I have tried to distinguish those), "does God exist?" Would that information have *any* relevant or importance in actually settling that question? Almost zero.

So, I want to emphasize that when I say "folk intuitions" are important, I only mean it in a very narrow context: in determining what a concept is, according to its common usage. Thus, if we did massive surveys and finally determined what the various concepts are, and we had a surprising amount of precision, then I would agree with you: don't give the folk these definitions, as we've determined them, and ask them whether compatibilism is true. I don’t trust the masses to get modus tollens right.

But even if this exception is narrow, it is still important. It blocks an escape route for compatibilists that otherwise find the prospects of defending their views to be bleak: “well, free will in that traditional, stronger sense isn’t compatible with determinism, but that isn’t what free will means or should mean.” And I think it is important to block that exit.

As for the rest of your post, I am afraid that, probably through no fault but my own, I did not fully understand all of what you wrote.

Paul,

I think the way you put your point is precisely the reason why I wouldn't say that desirability is an element of the meaning of `good' or `free will'. Perhaps being desirable is a way to identify these concepts, in the sense that most of us take `good' or `free will' to be desirable, but you have to ask yourself if this should lead us to conclude that desirability is part of the meaning of these concept?

I think that the only way to determine this is to look at common use as Kip has been claiming. Obviously common use is not a very clear concept, but whatever it is, I think that the test I proposed above does tell us something about common use. Once we can use a term such as `good' or `free will' with the adjective `undesirable' and this doesn't lead to a manifest contradiction, desirability is not part of the meaning of `good' or `free will' .

My relatively strict attention to this sort of inconsistency also leads me to believe that if anybody can truly prove that the folk concept of `free will' is internally inconsistent, such as the concept `nwhite' which means white and not-white (in the same spatio-temporal location), that there is then nothing left but a constructive move. For it is clear that there are no possible worlds in which something is `nwhite' and it is therefore a useless concept. In this constructive move I would agree with everything posted above about such moves, and would like to see a connection between the new concept of `free will'(*) and what we want this concept to do.

I would like to ad to this discussion by giving an example that I found in the literature I studied for my course (I'm not up to speed with the most contemporary literature) of the danger of confusing the conceptual analysis of `free will' as a merely descriptive activity and the activity of redefining `free will' so that it fits with compatibilism.

In `Freedom of the will and the concept of a person'', Frankfurt gives a critique of Chisholm for his version of agent causation, calling it quaint and claiming that Chisholm can't show why we would want free will or why we think that animals have no free will. Frankfurt, in this article, is clearly trying to redefine `free will' in a way such that it can meet these demands and perhaps a few others. However in his critique, he misses the fact that the point of Chisholm's position is to provide the metaphysics that fit with what he and I see as the true folk notion of `free will'. This notion includes something like the Forking Paths model or the Source model as described here http://plato.stanford.edu/entries/compatibilism/#2. Chisholm seems to have some faith in these metaphysics, but that is less relevant here. What is relevant is the fact that Frankfurt doesn't make an adequate distinction between conceptual analysis and constructive revision and that this leads him to give a misguided critique. It is because of these sorts of situations that it would be good if those who write about free will are clear about the type of activity they are engaging in.